The anatomy of a lost booking.

Friday, 7:42pm. Somewhere in your city, a couple has just decided to eat out. She messages one restaurant on WhatsApp. Rings a second; nobody picks up, because it's 7:42 on a Friday and service is loud. Opens a third's website and is asked to create an account before anyone will confirm a table even exists.

By 7:51 they're booked. Just not with any of those three.

Operators talk about losing bookings the way people talk about burglary: a dramatic event, a competitor climbing through the window. Years on the operator's side and six months building software for this industry have taught me it's almost never that. Bookings aren't stolen. They leak. Quietly, in five places, and in most venues nobody is watching the taps.

One: discovery. Before a guest can choose you, they have to find you, and finding has changed address. Guests increasingly ask an AI where to eat, and the AI answers from whatever it can actually read. If your menus, your rooms and your story live in a PDF and an unmonitored inbox, you aren't being rejected. You're not in the room. Try it tonight: ask ChatGPT for the best terrace near your own venue, and see who it names instead of you.

Two: the front door. The booking widget that demands an account before showing availability. Email, password, date of birth, marketing consent — all before anyone will admit whether Thursday at eight is possible. Hospitality is the only industry that runs a bouncer on the internet. The guest at 7:42 does not make an account. She makes other plans.

Three: the unanswered message. The WhatsApp from the top of this letter. Nobody ignored her; service doesn't stop for messages, and the person who could answer is carrying plates. But the guest can't see your Friday. She can only see the ticks. The unwatched inbox is the most expensive employee a venue never hired.

Four: the confirmation gap. Booking made Tuesday, dinner Saturday, silence in between. By Saturday the plan has wobbled, someone's tired, nobody re-committed them, and at 8:15 there's an empty table wearing a RESERVED sign. Venues file these under no-shows. In my experience a good half are better filed under never-quite-confirmed — which matters, because that half is preventable.

Five: the vanished regular. Came twice in March, loved it, told people. Never came back, and nobody noticed, because no system was counting. This is the quiet one — it doesn't cost you a table tonight, it costs you a customer forever, and it compounds.

I won't hand you a consultancy statistic about what all this costs; most of those numbers are marketing wearing a lab coat. Simple arithmetic instead. One leaked four-top a week is fifty-two dinners a year. Price them at your own average spend and sit with the number — then remember there are five taps, and that was one of them.

And I'll declare a scar of my own, because this letter would be dishonest without one. Last month I found our own widget showing a guest in London a Dubai venue as fully booked — on a date that, in Dubai, hadn't happened yet. The widget was reading the guest's clock instead of the venue's. We fixed it within days, but the lesson stands and it applies to everyone selling certainty in this industry, me included: every system leaks. The difference between venues isn't whether. It's whether anyone is counting.

Counting, as it happens, is most of what the AI Readiness Audit does when I run it on a venue — but that's a sentence, not this letter's subject.

Next week: what broke. Six months of plugging AI into real venues, told with the reverts left in.

— B

Disagree with any of this? Message me. I answer.

Bhrij Patel is the founder of RAYN, an AI-native hospitality intelligence platform, built after years operating restaurants from quick service to fine dining. For an honest, vendor-independent read on where AI fits your operation, ask about the AI Readiness Audit